PowderCoatingLogix/scripts/Web Scraping/backup/discover-prismatic-by-color-param.js

const fs = require("fs");
const { chromium } = require("playwright");

const headed = process.argv.includes("--headed");

function getArgValue(name, defaultValue) {
  const prefix = `--${name}=`;
  const found = process.argv.find(x => x.startsWith(prefix));
  return found ? found.slice(prefix.length) : defaultValue;
}

const maxScrollsPerColor = parseInt(getArgValue("max-scrolls-per-color", "180"), 10);
const stopAfterNoNewScrolls = parseInt(getArgValue("stop-after-no-new-scrolls", "10"), 10);

const baseUrl = "https://www.prismaticpowders.com/shop/powder-coating-colors";
const outputFile = "product-urls.txt";
const logFile = "color-discovery-log.json";

// Update this list if you find more color params in the site HTML.
const colorParams = [
  "pris_black",
  "pris_blue",
  "pris_bronze",
  "pris_brown",
  "pris_clear",
  "pris_copper",
  "pris_gold",
  "pris_gray",
  "pris_green",
  "pris_orange",
  "pris_pink",
  "pris_purple",
  "pris_red",
  "pris_silver",
  "pris_tan",
  "pris_white",
  "pris_yellow"
];

function cleanUrl(url) {
  return (url || "").split("?")[0].split("#")[0].trim();
}

function isProductUrl(url) {
  return /\/shop\/powder-coating-colors\/[A-Z0-9-]+\//i.test(url || "");
}

function readExistingUrls() {
  if (!fs.existsSync(outputFile)) return [];

  return fs.readFileSync(outputFile, "utf8")
    .split(/\r?\n/)
    .map(cleanUrl)
    .filter(Boolean);
}

function writeUrls(urls) {
  const sorted = [...urls].sort();
  fs.writeFileSync(outputFile, sorted.join("\r\n") + "\r\n", "utf8");
}

function readLog() {
  if (!fs.existsSync(logFile)) {
    return {
      completed_colors: {},
      runs: []
    };
  }

  try {
    return JSON.parse(fs.readFileSync(logFile, "utf8"));
  } catch {
    return {
      completed_colors: {},
      runs: []
    };
  }
}

function writeLog(log) {
  fs.writeFileSync(logFile, JSON.stringify(log, null, 2), "utf8");
}

async function collectProductLinks(page) {
  const links = await page.locator("a").evaluateAll(anchors =>
    anchors
      .map(a => a.href)
      .filter(Boolean)
      .filter(h => /\/shop\/powder-coating-colors\/[A-Z0-9-]+\//i.test(h))
  );

  return links.map(cleanUrl).filter(Boolean);
}

async function scrollAndCollect(page, urls, label) {
  let noNewScrolls = 0;
  let totalAddedForThisColor = 0;

  for (let i = 0; i < maxScrollsPerColor; i++) {
    const before = urls.size;

    for (const link of await collectProductLinks(page)) {
      urls.add(link);
    }

    const after = urls.size;
    const added = after - before;
    totalAddedForThisColor += added;

    if (added === 0) {
      noNewScrolls++;
    } else {
      noNewScrolls = 0;
    }

    writeUrls(urls);

    console.log(`[${label}] Scroll ${i + 1}/${maxScrollsPerColor}: +${added}, total ${after}, no-new ${noNewScrolls}`);

    if (noNewScrolls >= stopAfterNoNewScrolls) {
      break;
    }

    await page.mouse.wheel(0, 2500);
    await page.waitForTimeout(1500);
  }

  return totalAddedForThisColor;
}

(async () => {
  const existingUrls = readExistingUrls();
  const urls = new Set(existingUrls);
  const log = readLog();

  console.log(`Existing URLs in ${outputFile}: ${existingUrls.length}`);

  const browser = await chromium.launch({ headless: !headed });

  const context = await browser.newContext({
    userAgent: "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36",
    viewport: { width: 1365, height: 900 },
    locale: "en-US",
    timezoneId: "America/New_York"
  });

  const page = await context.newPage();

  const runRecord = {
    started_at: new Date().toISOString(),
    existing_at_start: existingUrls.length,
    colors_attempted: []
  };

  for (const color of colorParams) {
    if (log.completed_colors[color]) {
      console.log(`Skipping completed color: ${color}`);
      continue;
    }

    const url = `${baseUrl}?color=${encodeURIComponent(color)}`;
    console.log("");
    console.log(`Opening color filter: ${color}`);
    console.log(url);

    try {
      const response = await page.goto(url, {
        waitUntil: "domcontentloaded",
        timeout: 60000
      });

      const status = response ? response.status() : "unknown";
      console.log(`HTTP status: ${status}`);

      await page.waitForTimeout(5000);

      const before = urls.size;
      const addedDuringScroll = await scrollAndCollect(page, urls, color);
      const after = urls.size;
      const netAdded = after - before;

      log.completed_colors[color] = {
        url,
        http_status: status,
        added: netAdded,
        added_during_scroll: addedDuringScroll,
        total_after: after,
        completed_at: new Date().toISOString()
      };

      runRecord.colors_attempted.push({
        color,
        url,
        http_status: status,
        added: netAdded,
        total_after: after
      });

      writeLog(log);
      writeUrls(urls);

      console.log(`Color complete: ${color}; added ${netAdded}; total ${after}`);

      // Polite pause between filters.
      await page.waitForTimeout(3000);
    } catch (err) {
      console.log(`Color failed: ${color}; ${err.message}`);

      runRecord.colors_attempted.push({
        color,
        url,
        added: 0,
        error: err.message
      });

      writeLog(log);
    }
  }

  runRecord.finished_at = new Date().toISOString();
  runRecord.final_total = urls.size;
  runRecord.new_this_run = urls.size - existingUrls.length;

  log.runs.push(runRecord);
  writeLog(log);
  writeUrls(urls);

  console.log("");
  console.log("Color-param discovery complete.");
  console.log(`Existing at start: ${existingUrls.length}`);
  console.log(`Final total: ${urls.size}`);
  console.log(`New this run: ${urls.size - existingUrls.length}`);
  console.log(`Output: ${outputFile}`);
  console.log(`Log: ${logFile}`);

  await browser.close();
})();